CHAPTER VIII.

GENERAL ACCOUNT OF THE SKELETON IN FISHES[39].

EXOSKELETON.

The most primitive type of exoskeleton is that found in Elasmobranchs and formed of =placoid= scales; these are tooth-like structures consisting of dentine and bone capped with enamel, and have been already described (p. 4). In most Elasmobranchs they are small and their distribution is fairly uniform, but in the Thornback skate, _Raia clavata_, they have the form of larger, more scattered spines. In adult Holocephali and in _Polyodon_ and _Torpedo_ there is no exoskeleton, in young Holocephali, however, there are a few small dorsal ossifications.

The plates or scales of many Ganoids may have been formed by the gradual fusion of elements similar to these placoid scales, and often bear a number of little tooth-like processes. In _Lepidosteus_, _Polypterus_, and many extinct species, these _ganoid_ scales, which are rhomboidal in form and united to one another by a peg and socket articulation, enclose the body in a complete armour. In _Trissolepis_ part of the tail is covered by rhomboidal scales, while rounded scales cover the trunk and remainder of the tail. _Acipenser_ and _Scaphirhynchus_ have large dermal bony plates which are not rhomboidal in shape and do not cover the whole body. In _Acipenser_ a single row extends along the middle of the back and two along each side.

The majority of Teleosteans have thin flattened scales which differ from those of Ganoids in being entirely mesodermal in origin, containing no enamel. There are two principal types of Teleostean scales, the cycloid and ctenoid. A =cycloid= scale is a flat thin scale with concentric markings and an entire posterior margin. A =ctenoid= scale differs in having its posterior margin pectinate. The Dipnoi have overlapping cycloid scales. The rounded scales of _Amia_ and of many fossil ganoids such as _Holoptychius_ are shaped like cycloid scales, but differ from them in being more or less coated with enamel. In Eels and some other Teleosteans the scales are completely degenerate and have almost disappeared. Some Teleosteans, like _Diodon hystrix_, have scales with triradiate roots from which arise long sharp spines directed backwards. These scales, which resemble teeth, contain no enamel; they become erect when the fish inflates its body into a globular form. Many Siluroids have dermal armour in the form of large bony plates which are confined to the anterior part of the body. In _Ostracion_ the whole body is covered by hexagonal plates, closely united together.

The =fin-rays= are structures of dermal origin which entirely or partially support the unpaired fins, and assist the bony or cartilaginous endoskeleton in the support of the paired fins.

In Elasmobranchs, Dipnoi, and Chondrosteous ganoids the skeletons of the fins are, as a rule, about half of exoskeletal, half of endoskeletal origin, the proximal and inner portion being cartilaginous and endoskeletal, the distal and outer portion being exoskeletal, and consisting of horny or of more or less calcified fin-rays. In bony Ganoids and Teleosteans the endoskeletal parts are greatly reduced and the fins come to consist mainly of the fin-rays, which are ossified and frequently become flattened at their distal ends.

The fin-rays of the ventral part of the caudal fin are carried by the haemal arches; those of the dorsal and anal fins and of the dorsal part of the caudal fin generally by interspinous bones, which in adult Teleosteans alternate with the neural and haemal spines. In Dipnoi these interspinous bones articulate with the neural and haemal spines. In many Siluroids the anterior rays of the dorsal and pectoral fins are developed into large spines which often articulate with the endoskeleton, or are sometimes fused with the dermal armour plates. Similar spines may occur in Ganoids in front of both the dorsal and anal fins. _Polypterus_ has a small spine or _fulcrum_ in front of each segment of the dorsal fin. Such spines are often found fossilised, and are known as _ichthyodorulites_.

Similar spines are found in many Elasmobranchs, but they are simply inserted in the flesh, not articulated to the endoskeleton. They also differ from the spines of Teleosteans and Ganoids in the fact that they are covered with enamel, and often have their edges serrated like teeth. In the extinct Acanthodii they generally occur in front of all the fins, paired and unpaired.

In _Trygon_, the Sting-ray, the tail bears a serrated spine which is used for purposes of offence and defence. Many ichthyodorulites may have been spines of this nature fixed to the tail, rather than spines situated in front of the fins. The spines, which are always found in front of the dorsal fin in Holocephali, agree with those of Elasmobranchs in containing enamel, and with those of Teleosteans in being articulated to the endoskeleton.

TEETH.

The teeth of fish[40] are subject to a very large amount of variation, perhaps to more variation than are those of any other class of animals. Sometimes, as in adult Sturgeons, they are entirely absent, sometimes they are found on all the bones of the mouth, and also on the hyoid and branchial arches. The teeth are all originally developed in the mucous membrane of the mouth, but they afterwards generally become attached to firmer structures, especially to the jaws. In Elasmobranchs, however, they are generally simply imbedded in the tough fibrous integument of the mouth. Their attachment to the jaws may take place in three different ways.

(1) By an elastic hinge-joint, as in the Angler (_Lophius_), and the Pike (_Esox lucius_). In the Angler the tooth is held by a fibrous band attaching its posterior end to the subjacent bone, in the Pike by uncalcified elastic rods in the pulp cavity.

(2) By ankylosis, i.e. by the complete union of the calcified tooth substance with the subjacent bone. This is the commonest method among fish.

(3) By implantation in sockets. This method is not very common among fish. The teeth are sometimes, as in _Lepidosteus_, ankylosed to the base of the socket. In this genus there is along each ramus of the mandible a median row of large teeth placed in perfect sockets, and two irregular lateral rows of small teeth ankylosed to the jaw.

Dentine, enamel and cement are all represented in the teeth of fishes, but the enamel is generally very thin, and cement is but rarely developed. Dentine forms the main bulk of the teeth; it is sometimes of the normal type, but generally differs from that in higher vertebrates in being vascular, and is known as _vasodentine_. A third type occurs, known as _osteodentine_; it is traversed by canals occupied by marrow, and is closely allied to bone.

The teeth are generally continually renewed throughout life, but sometimes one set persists.

The teeth of Selachii are fundamentally identical with placoid scales. They are developed from a layer of dental germs which occurs all over the surface of the skin, except in the region of the lips. At this point the layer of tooth-producing germs extends back into the mouth, being projected by a fold of the mucous membrane (fig. 14, 7). Here new teeth are successively formed, and as they grow each is gradually brought into a position to take the place of its predecessor by the shifting outwards of the gum over the jaw. Owing to this arrangement sharks have practically an unlimited supply of teeth (figs. 14 and 15).

Two principal types of teeth are found in ELASMOBRANCHS. In Sharks and Dogfish, on the one hand, the teeth are very numerous, simple, and sharp-pointed, and are with or without serrations and lateral cusps. Many Rays and fossil Elasmobranchs, on the other hand, have broad flattened teeth adapted for crushing shells. Intermediate conditions occur between these two extremes. Thus in _Cestracion_ and many extinct sharks, such as _Acrodus_, while the median teeth are sharp, the lateral teeth are more or less flattened and adapted for crushing. In various species belonging to the genus _Raia_ the teeth of the male are sharp, while those of the female are blunt. A very specialised dentition is met with in the Eagle-rays (Myliobatidae), in which the jaws are armed with flattened angular tooth-plates, arranged in seven rows, forming a compact pavement; the plates of the middle row are very wide and rectangular, those of the other rows are much smaller and hexagonal. Lastly, in _Cochliodus_ the individual crushing teeth are fused, forming two pairs of spirally-coiled dental plates on each side of each jaw. _Pristis_, the Saw-fish, has a long flat cartilaginous snout, bearing a double row of persistently-growing teeth planted in sockets along its sides. Each tooth consists of a number of parallel dentinal columns, united at the base, but elsewhere distinct.

In the HOLOCEPHALI--_Chimaera_, _Hariotta_ and _Callorhynchus_--only three pairs of teeth or dental plates occur, two pairs in the upper jaw, one in the lower. These structures persist throughout life and grow continuously. The upper tooth structures are attached respectively to the ethmoid or vomerine region of the skull, and to the palato-pterygoids. The vomerine teeth are small, while those attached to the mandible and the palato-pterygoid region are large and bear several roughened ridges adapted for grinding food. The teeth of the two opposite sides of the jaw meet in a median symphysis. The teeth of _Chimaera_ are more adapted for cutting, those of _Callorhynchus_ for crushing. Many extinct forms are known, some of whose teeth are intermediate in structure between those of _Chimaera_ and _Callorhynchus_.

The teeth of GANOIDS are also extremely variable. Among living forms, the Holostei are more richly provided with teeth than are any other fishes, as they may occur on the premaxillae, maxillae, palatines, pterygoids, parasphenoid, vomers, dentaries, and splenials. Among the Chondrostei, on the other hand, the adult Acipenseridae are toothless; small teeth however occur in the larval sturgeon, and in _Polyodon_ many small teeth are found attached merely to the mucous membrane of the jaws. Many fossil Ganoids have numerous flattened or knob-like teeth, borne on the maxillae, palatines, vomers and dentaries. Others have a distinctly heterodont dentition. Thus in _Lepidotus_ the premaxillae bear chisel-like teeth, while knob-like teeth occur on the maxillae, palatines and vomers. In _Rhizodus_ all the teeth are pointed, but while the majority are small a few very large ones are interspersed.

In TELEOSTEANS, too, the teeth are eminently variable both in form and mode of arrangement. They may be simple and isolated, or compound, and may be borne on almost any of the bones bounding the mouth cavity, and also as in the Pike, on the hyoid and branchial arches. The splenial however never bears teeth and the pterygoid and parasphenoid only rarely, thus differing from the arrangement in the Holostei.

The isolated teeth are generally conical in form and are ankylosed to the bone that bears them. Such teeth are, with a few exceptions such as _Balistes_, not imbedded in sockets nor replaced vertically.

In some fish beak-like structures occur, formed partly of teeth, partly of the underlying jaw bones. These beaks are of two kinds: (1) In _Scarus_, the parrot fish, the premaxillae and dentaries bear numerous small, separately developed teeth, which are closely packed together and attached by their proximal ends to the bone, while their distal ends form a mosaic. Not only the teeth but the jaws which bear them are gradually worn away at the margins, while both grow continuously along their attached edge. (2) In Gymnodonts, e.g. _Diodon_, the beaks are formed by the coalescence of broad calcified horizontal plates, which when young are free and separated from one another by a considerable interval.

In some Teleosteans the differentiation of the teeth into biting teeth and crushing teeth is as complete as in _Lepidosteus_. Thus in the Wrasse (_Labrus_), the jaws bear conical slightly recurved teeth arranged in one or two rows, with some of the anterior ones much larger than the rest. The bones of the palate are toothless, while both upper and lower pharyngeal bones are paved with knob-like crushing teeth; such pharyngeal teeth occur also in the Carp but are attached only to the lower pharyngeal bone, the jaw bones proper being toothless.

In DIPNOI the arrangement of the teeth is very similar to that in Holocephali. The mandible bears a single pair of grinding teeth attached to the splenials, and a corresponding pair occur on the palato-pterygoids. In front of these there are a pair of small conical vomerine teeth loosely attached to the ethmoid cartilage. The palato-pterygoid teeth of _Ceratodus_ are roughly semicircular in shape with a smooth convex inner border, and an outer border bearing a number of strongly marked ridges. The teeth of the extinct Dipteridae resemble those of _Ceratodus_ but are more complicated.

ENDOSKELETON.

SPINAL COLUMN[41].

The spinal column of fishes is divisible into only two regions, a caudal region in which the haemal arches or ribs meet one another ventrally, and a precaudal region in which they do not meet.

The various modifications of the spinal column in fishes can be best understood by comparing them with the arrangement in the simplest type known, namely _Amphioxus_. In _Amphioxus_ the notochord is immediately surrounded by a structureless cuticular layer, the _chordal sheath_. Outside this is the _skeletogenous layer_, which in addition to surrounding the notochord and chordal sheath embraces the nerve cord dorsally, and laterally sends out septa forming the _myomeres_.

The CARTILAGINOUS GANOIDS[42] _Acipenser_, _Polyodon_ and _Scaphirhynchus_ are the simplest fishes as regards their spinal column. The notochord remains permanently unconstricted and is enclosed in a chordal sheath, external to which is the skeletogenous layer. In this layer the development of cartilaginous elements has taken place. In connection with each _neuromere_, or segment as determined by the points of exit of the spinal nerves, there are developed two pairs of ventral cartilages, the ventral arches (basiventralia) and intercalary pieces (interventralia); and at least two pairs of dorsal pieces, the neural arches (basidorsalia) and intercalary pieces (interdorsalia). The lateral parts of the skeletogenous layer do not become converted into cartilage, so there are no traces of vertebral centra. The ventral or haemal arches meet one another ventrally and send out processes to protect the ventral vessels. The neural arches do not meet, but are united by a longitudinal elastic band.

In Cartilaginous ganoids the only indications of metameric segmentation are found in the neural and haemal arches. The case is somewhat similar with the Holocephali and Dipnoi.

In the HOLOCEPHALI the notochord grows persistently throughout life, and is of uniform diameter throughout the whole body except in the cervical region and in the gradually tapering tail. The chordal sheath is very thick and includes a well-marked zone of calcification which separates an outer zone of hyaline cartilage from an inner zone. There are also a number of cartilaginous pieces derived from the skeletogenous layer which are arranged in two series, a dorsal series forming the neural arches and a ventral series forming the haemal arches. These do not, except in the cervical region, meet one another laterally round the notochord and form centra. To each neuromere there occur a pair of basidorsals, a pair of interdorsals, and one or two supradorsals. In the tail the arrangement is irregular.

In the DIPNOI as in the Holocephali the notochord grows persistently and uniformly, and the chordal sheath is thick and cartilaginous though there are no metamerically arranged centra. The neural and haemal arches and spines are cartilaginous and interbasalia (intercalary pieces) are present. The basidorsalia and basiventralia do not in _Ceratodus_ meet round the notochord and enclose it except in the anterior part of the cervical and posterior part of the caudal region.

In ELASMOBRANCHII the chordal sheath is weak and the skeletogenous layer strong. Biconcave cartilaginous vertebrae are developed, and as is the case in most fishes, constrict the notochord _vertebrally_.

Two distinct types of vertebral column can be distinguished in Elasmobranchs[43]:

1. In many extinct forms and in the living Notidanidae, _Cestracion_, and _Squatina_, the dorsal and ventral arches do not meet one another laterally round the centrum, and consequently readily come away from it.

2. In most living Elasmobranchs the arches meet laterally round the centrum.

The vertebrae are never ossified but endochondral calcification nearly always takes place, though it very rarely reaches the outer surface of the vertebrae. Elasmobranchs are sometimes subdivided into three groups according to the method in which this calcification takes place:

1. =Cyclospondyli= (_Scymnus_, _Acanthias_), in which the calcified matter is deposited as one ring in each vertebra.

2. =Tectospondyli= (_Squatina_, _Raia_, _Trygon_), in which there are several concentric rings of calcification.

3. =Asterospondyli= (Notidanidae, _Scyllium_, _Cestracion_), in which the calcified material instead of forming one simple ring, extends out in a more or less star-shaped manner.

In _Heptanchus_ the length of the vertebral centra in the middle of the trunk is double that in the anterior and posterior portions, and as the length of the arches does not vary, the long centra carry more of them than do the short centra.

In many Rays the skull articulates with the vertebral column by distinct occipital condyles.

In BONY GANOIDS the skeletogenous layer becomes calcified ectochondrally in such a way that the notochord is pinched in at intervals, and distinct vertebrae are produced. Ossification of the calcified cartilage rapidly follows. In _Amia_ the vertebrae are biconcave, in _Lepidosteus_ they are opisthocoelous, cup and ball joints being developed between the vertebrae in a manner unique among fishes. The notochord entirely disappears in the adult _Lepidosteus_, but at one stage in larval life it is expanded vertebrally and constricted intervertebrally in the manner usual in the higher vertebrata, but unknown elsewhere among fishes.

The tail of _Amia_ is remarkable from the fact that as a rule to each neuromere, as determined by the exit of the spinal nerves, there are two centra, a posterior one which bears nothing, and an anterior one which bears the neural and haemal arches, these being throughout the vertebral column connected with the centra by cartilaginous discs.

In most TELEOSTEANS but not in the Plectognathi the neural arches are continuous with the centra, which are nearly always deeply biconcave.

In some cases many of the anterior vertebrae are ankylosed together and to the skull. The vertebrae often articulate with one another by means of obliquely placed flattened surfaces, the zygapophyses. The centrum in early stages of development is partially cartilaginous, but the neural arches and spines in the trunk at any rate, pass directly from the membranous to the osseous condition.

FINS.

The most primitive fins are undoubtedly the unpaired ones, which probably originally arose as ridges or folds of skin along the mid-dorsal line of the body, and passed thence round the posterior end on to the ventral surface, partially corresponding in position and function to the keel of a ship.

In long 'fish' which pass through the water with an undulating motion such simple continuous fins may be the only ones found, as in _Myxine_. To support these median fins skeletal structures came to be developed; these show two very distinct forms, viz. cartilaginous endoskeletal pieces, the _radiale_, and horny exoskeletal fibres, the _fin-rays_. Mechanical reasons caused the fin to become concentrated at certain points and reduced at intervening regions. Thus a terminal caudal fin arose and became the chief organ of propulsion, and the dorsal and ventral fins became specialised to act as balancing organs.

In some of the earlier Elasmobranchs, the Pleuracanthidae, the endoskeletal cartilaginous radiale are directly continuous with outgrowths from the dorsal and ventral arches of the vertebrae, and form the main part of the fin. In later types of Elasmobranchs the horny exoskeletal fin-rays have comparatively greater prominence. In bony fish, as has been already stated, the horny fibres are replaced by bony rays of dermal origin, and at the same time complete reduction and disappearance of the cartilaginous radiale takes place.

THE CAUDAL FIN.

The caudal region of the spinal column in fishes is of special importance. It is distinctly marked off from the rest of the spinal column by the fact that the ventral or haemal arches meet one another and are commonly prolonged into spines, while in the trunk region they do not meet but commonly diverge from one another.

In some fish the terminal part of the caudal region of the spinal column retains the same direction as the rest of the spinal column. The blade of the caudal fin is then divided into two nearly equal portions, and is said to be =diphycercal=. This condition is generally regarded as the most primitive one; it occurs in the Ichthyotomi, Holocephali, all living Dipnoi, _Polypterus_ and some extinct Crossopterygii, and a few Selachii and Teleostei. It occurs also in deep-sea fish belonging to almost every group, and under these conditions obviously cannot be regarded as primitive, but must be looked on as a feature induced by the peculiar conditions of life.

In the great majority of fish the terminal part of the caudal region of the spinal column is bent dorsalwards, and the part of the blade of the caudal fin which arises on the dorsal surface is much smaller than is that arising on the ventral surface. Such a fin is said to be =heterocercal=.

Strictly speaking all fish whose tails are not diphycercal have heterocercal tails, but the term is commonly applied to two-bladed tails in which the spinal column forms a definite axis running through the dorsal blade, while the ventral blade is enlarged and generally forms the functional part of the tail. Such heterocercal tails are found in nearly all Elasmobranchii, together with the living cartilaginous Ganoidei, and many extinct forms belonging to the same order; _Lepidosteus_, _Amia_, and the Dipteridae among Dipnoi, have tails which, though obviously heterocercal, are not two-bladed.

The vast majority of the Teleostei and some extinct Ganoidei have heterocercal tails of the modified type to which the term =homocercal= is applied. The hypural bones which support the lower half of the tail fin become much enlarged, and frequently unite to form a wedge-shaped bone which becomes ankylosed to the last ossified vertebral centrum. The fin-rays then become arranged in such a way as to produce a secondary appearance of symmetry. Some homocercal fish such as the Perch have the end of the notochord protected by a calcified or completely ossified sheath, the =urostyle=, to which several neural and haemal arches may be attached, and which becomes united with the centrum of the last vertebra; in others such as the Salmon the end of the notochord is protected only by laterally placed bony plates.

THE SKULL.

It is often impossible to draw a hard and fast line between the cranium and the vertebral column. This is the case for instance in _Acipenser_ (fig. 18, 16) among Chondrostei, in _Amia_ among Holostei, and in _Ceratodus_ and _Protopterus_ among Dipnoi. The occipital region of the skull in _Amia_ is clearly formed of three cervical vertebrae whose centra have become absorbed into the cranium, while the neural arches and spines are still distinguishable.

The simplest type of cranium is that found in ELASMOBRANCHS: it consists of a simple cartilaginous box, which is generally immovably fixed to the vertebral column, though in some forms, like _Scymnus_ and _Galeus_, a joint is indicated, and in others, such as the Rays, one is fairly well developed. The cranium in Elasmobranchs is never bony, though the cartilage is sometimes calcified. It is drawn out laterally into an antorbital process in front of the eye, and a postorbital process behind it. The nasal capsules are always cartilaginous, and the eye, as a general rule, has a cartilaginous sclerotic investment. The cranium is often prolonged in front into a rostrum which is enormously developed in _Pristis_ and some Rays. The cartilaginous roof of the cranium is rendered incomplete by the presence of a large hole, the anterior fontanelle.

Two pairs of labial cartilages (fig. 16, B, 8) are often present. They lie imbedded in the cheeks outside the anterior region of the jaws, and are specially large in _Squatina_.

As regards the visceral arches[44] the simplest and most primitive condition of the jaws is that of the Notidanidae, in which the mandibular and hyoid arches are entirely separate. In these primitive fishes the palato-pterygo-quadrate bar articulates with the postorbital process (fig. 16, 10), while further forwards it is united to the cranium by the ethmo-palatine ligament. The hyoid arch is small and is broadly overlapped by the mandibular arch. The term =autostylic= is used to describe this condition of the suspensorium. From this condition we pass in the one direction to that of _Cestracion_ (fig. 16, B), in which the whole of the palato-pterygo quadrate bar has become bound to the cranium, and in the other to that of _Scyllium_. In _Scyllium_ (fig. 6), while the ethmo-palatine ligament is retained, the postorbital articulation of the palato pterygo-quadrate has been given up, so that the palato-pterygo quadrate comes to abut on the hyomandibular and is attached to it by ligaments. The pre-spiracular ligament (fig. 16, 20) running from the auditory capsule also assists in supporting the jaws.

Lastly we come to the purely =hyostylic= condition met with in Rays, in which the mandibular arch is entirely supported by the hyomandibular. In some Rays the hyoid is attached to the posterior face of the hyomandibular near its proximal end, and may even come to articulate with the cranium.

The =visceral arches of Elasmobranchs= may be summarised as follows:--

1. The =mandibular arch=, consisting of a much reduced dorsal portion, the pre-spiracular ligament, and a greatly developed ventral portion from which both upper and lower jaws are derived. The mandible (Meckel's cartilage) is the original lower member of the mandibular arch, and from it arises an outgrowth which forms the upper jaw or palato-pterygo-quadrate bar. In _Scymnus_ this bears a few branchiostegal rays.

2. The =hyoid arch=, which consists of the hyomandibular and the hyoid, and bears branchiostegal rays on its posterior face.

3. The =branchial arches=, generally five in number, all of which except the last bear branchiostegal rays. In the Notidanidae the number of branchial arches is increased beyond the normal series, thus in _Hexanchus_ there are six, and in _Heptanchus_ seven. There are six also in _Chlamydoselache_ and _Protopterus_.

4. The so-called external branchial arches which are cartilaginous rods attached to all the visceral arches. They are especially large in _Cestracion_.

The skull in HOLOCEPHALI is entirely cartilaginous. The palato-pterygo-quadrate bar is fixed to the cranium, and to it the mandible articulates. There is a well-marked joint between the skull and the spinal column.

In living Cartilaginous GANOIDS the primitive cartilaginous cranium is very massive, and is greatly prolonged anteriorly, while posteriorly it merges into the spinal column. Although it is mainly cartilaginous a number of ossifications take place in the skull, and membrane bones are now found definitely developed, especially in connection with the roof of the cranium. In _Acipenser_ (fig. 18) the ossifications in the cartilage include the pro-otic, which is pierced by the foramen for the fifth nerve, the alisphenoid, orbitosphenoid, ectethmoid, palatine, pterygoid, meso-pterygoid, hyomandibular (fig. 18, 11), cerato-hyal, all the cerato-branchials, and the first two epi-branchials. Most of these structures are, however, partly cartilaginous, though they include an ossified area. The membrane bones too of _Acipenser_ are very well developed, they include a bone occupying the position of the supra-occipital, and form a complete dorsal cephalic shield. Resting on the ventral surface are a vomer and a very large parasphenoid (fig. 18, 3). There is a bony operculum attached to the hyomandibular, and membrane bones representing respectively the maxilla and dentary are attached to the jaws. The suspensorium is most markedly hyostylic. The palato-pterygo-quadrate bar has a very curious shape and is quite separate from the cranium. It is connected to the hyomandibular by a thick symplectic ligament containing a small bone homologous with the symplectic of Teleosteans.

_Polyodon_ differs much from _Acipenser_, the membrane bones not being so well developed though they cover the great cartilaginous snout.

The skull in _Polypterus_ (Crossopterygii) shows a great advance towards the condition met with in Teleostei. The cranium remains to a great extent unossified, and large dorsal and ventral fontanelles pierce its walls. It is covered by a great development of membrane bones, paired nasals, frontals, parietals, supra- and post-temporals, and dermo-supra-occipitals among others being present. The palato-pterygo-quadrate bar is fused to the cranium, and in connection with it the following paired membrane bones appear, palatine, ecto-, meso- and meta-pterygoid, and further forwards jugal, vomer, maxilla and premaxillae. The membrane bones developed in connection with each ramus of the mandible are the dentary, angular, and splenial, in addition to the cartilage bone the articular. Several large opercular bones occur. There are also a pair of large jugular or gular plates, and several large opercular bones.

In Bony Ganoids both cartilage bone and membrane bone is well developed. The pro-otics and exoccipitals are well ossified, but the supra-occipital and pterotics are not. Lateral ethmoids are developed, and there are ossifications in the sphenoidal region which vary in different forms. The place of the cartilaginous palato-pterygo quadrate is taken by a series of bones, the quadrate behind and the palatine, ecto-, meso- and meta-pterygoids in front. In _Lepidosteus_, however, the palatine and pterygoid are membrane bones, as they are in _Polypterus_ and the Frog. Paired maxillae, premaxillae, vomers and a parasphenoid occur forming the upper jaw and roof of the mouth, and a series of membrane bones are found investing the mandible and forming the operculum.

In _Amia_[45] membrane bones are as freely developed as they are in Teleosteans; they include on each side a squamosal, four opercular bones, a lachrymal, a pre-orbital, one or two suborbitals, two large postorbitals and a supratemporal; while investing the mandible, besides the dentary, splenial, angular, and supra-angular, there is an unpaired jugular. The articular too is double and a mento-meckelian occurs. In _Amia_ teeth are borne on the premaxillae, maxillae, vomers, palatines and pterygoids.

Bony Ganoids are the lowest animals in which squamosal bones are found, and they do not occur in Teleosteans.

The suspensorium in bony Ganoids, as in the Chondrostei, is hyostylic, and there are two ossifications in the hyomandibular cartilage, viz. the hyomandibular, and the symplectic.

The skull of TELEOSTEI is very similar to those of _Lepidosteus_ and _Amia_. Although the bony skull is greatly developed and very complicated, much of the original cartilaginous cranium often persists. Membrane bones are specially developed on the roof of the skull where they include the parietal, frontal, and nasal bones. The same bones are developed in connection with the upper jaw and roof of the mouth as in bony Ganoids, but only two membrane bones occur in the lower jaw, viz. the angular and dentary. A number of large ossifications take place in the cartilage of the auditory capsules. In some forms parts of the last pair of branchial arches are broadened out and form the pharyngeal bones which bear teeth. The opercular bones and those of the upper and lower jaws are quite comparable to those of bony Ganoids.

A full account of the Teleostean skull has been given in the case of the Salmon (pp. 87-96) and the Cod (pp. 96-101).

In DIPNOI the skull is mainly cartilaginous, but both cartilage- and membrane-bone occur also. Cartilage-bone is found in the ossified exoccipitals, while of membrane-bones _Protopterus_ has among unpaired bones a fronto-parietal, a median ethmoid, and a parasphenoid, and among paired bones nasals and large supra-orbitals. The skull of _Ceratodus_ (fig. 19) has an almost complete roof of membrane bones, including some whose homology is doubtful. The ethmo-vomerine region is always cartilaginous, but bears small teeth. The palato-pterygo quadrate bar is ossified and firmly united to the cranium, and the mandible articulates directly with it (autostylic). Membrane bones are freely developed in connection with the mandible, dentary, splenial, and angular bones being all present. There are two opercular bones.

In the extinct Dipteridae the cranium is very completely covered with plates of dermal bone, and the skeleton in general is more ossified than is the case in recent Dipnoi.

Six pairs of branchial arches occur in _Protopterus_; _Ceratodus_ and _Lepidosiren_ have five, like most other fish. The branchial arches bear gill rakers.

RIBS.

As has been already mentioned (p. 24), although ribs commonly appear to be the cut-off ends of the transverse processes, they are really elements derived from the ventral or haemal arch.

In Elasmobranchii and other cartilaginous fish they have the form of small cartilaginous structures imperfectly separated from the diverging halves of the ventral arch, and are often absent.

In Teleostei and bony Ganoids they often have different attachments in different parts of the body. In the tail region they are not differentiated from the two halves of the ventral arch, which meet in the middle line, and are prolonged into a haemal spine. In the posterior trunk region they sometimes form distinct processes diverging from the two halves of the ventral arch; while further forward they may shift their attachment so as to arise from the dorsal side of the two halves of the ventral arch and at some distance from their ends, which now diverge as ventri-lateral processes.

APPENDICULAR SKELETON.

PECTORAL GIRDLE.

The simplest type of pectoral girdle is found in Elasmobranchs. It is entirely cartilaginous and consists of a curved ventrally-placed rod, ending dorsally in two horn-like scapular processes which are sometimes attached to the cranium or vertebral column. In Rays the shoulder girdle is very large, and has a distinct suprascapular portion forming a broad plate attached to the neural spines of the vertebrae. There is often a cup-like glenoid cavity for the articulation of the limb; this cavity is specially large in Rays and is much pierced by holes. In Dipnoi the cartilaginous girdle still occurs, but on it there is a deposit of membrane bone forming the clavicle, infraclavicle, and supra-clavicle. These bones, which with the exception of the clavicle, are unknown in higher vertebrates, are better developed in Ganoids, and best of all in Teleosteans. They are connected by the supratemporal with the epi-otic and opisthotic regions of the cranium. Owing to this development of dermal bone, the original cartilaginous arch becomes much reduced, but ossifications representing the scapula and coracoid occur in bony Ganoids and Teleosteans.

PELVIC GIRDLE.

In Elasmobranchs the pelvic girdle consists of a short ventral rod of cartilage representing the ischium and pubis, which does not send up dorsal iliac processes. In _Chimaera_ the pelvic girdle has a flattened pointed iliac portion, and ventrally an unpaired movable cartilaginous plate which bears hooks and is supposed to be copulatory in function. Claspers of the usual type are present as well. The Dipnoi have a primitive kind of pelvis in the form of a cartilaginous plate lying in the mid-ventral line and drawn out into three horns anteriorly. In Ganoids the pelvis has almost entirely disappeared, though small cartilaginous vestiges of it remain in _Polypterus_. In Teleosteans even these vestiges are gone, and in these fish and Ganoids the place of the pelvis is taken by the enlarged basi-pterygia (meta-pterygia) of the fins.

PAIRED FINS[46].

As regards the origin of the limbs or paired fins of fishes there are two principal views. One view, that of Gegenbaur, considers that limbs and their girdles are derived from visceral arches which have migrated backwards. The other view, which probably now has the greater number of supporters, considers that the paired fins of fishes are of essentially the same nature as the median fins.

According to Gegenbaur's view[47] the =archipterygium= of _Ceratodus_ (fig. 20) represents the lowest type of fin; it consists of a central cartilaginous axis bearing a large number of radiale. The dorsal or pre-axial radiale are more numerous than the ventral or postaxial, and at the margin of the fin[48] the cartilaginous endoskeletal radiale are replaced by horny exoskeletal fin-rays.

It is impossible here to give a full discussion of the rival views, but some of the points which support Gegenbaur's view may be mentioned. The fact that migration of visceral arches has to be assumed is no difficulty, as it is obvious that migration in the opposite direction has taken place in many Teleosteans such as the Cod, whose pelvic fins are attached to the throat in front of the pectorals. If migration did take place, the pelvic fins being older than the pectoral should be the more modified, and this is the case. Again, if the pectoral girdle is a modified branchial arch, it must at some period have carried a gill, and in _Protopterus_ it does bear a vestigial gill.

According to the view more prevalent at the present time, the paired fins have been derived from two continuous folds of skin and their skeletal supports running forward from the anal region along the sides of the body, their character being similar to the fold that gave rise to the median fins. In support of this view it may be argued that the paired and unpaired fins are often identical in structure, and that some Elasmobranch embryos do show a ridge running between the pectoral and pelvic fins. Then from this continuous fold two pairs of smaller folds may have been specialised off, and in each a number of cartilaginous radiale may have been developed. The fin of _Cladoselache_ from the Carboniferous of Ohio apparently illustrates this condition. It consists of certain basal pieces which do not project beyond the body wall and bear a number of unsegmented cartilaginous radiale, which show crowding together and are sometimes bifurcated distally; they extend throughout the whole fin from the body wall to the margin. From this fin the archipterygium might be easily derived by the enlargement of one of the middle radiale and the segmentation and partial fusion of them all.

Whether the archipterygium be a primitive or secondary type of fin, when it is once reached it is easy to derive all the other types from it. The fins of the other living Dipnoi,--_Protopterus_ and _Lepidosiren_--are simply archipterygia from which the radiale have almost or completely disappeared, leaving only the segmented axes. Archipterygia too are found in the pectoral fins of the Ichthyotomi, but the postaxial radiale are much reduced.

The =ichthyopterygium=, or type of fin, characteristic of many modern Elasmobranchs such as _Scyllium_, may have been derived from the archipterygium by the gradual reduction of the rays on the postaxial side of the axis and their condensation on the pre-axial side. The Ichthyotomi such as _Xenacanthus_ show one stage in the reduction of the postaxial rays, and a further stage is seen in the Notidanidae and some other sharks like _Scymnus_ and _Acanthias_, in which a few postaxial rays still remain. The condensation of the pre-axial rays when further continued leads to one of the rays getting an attachment to the girdle. Thus the fin comes to articulate with the girdle by two basalia or basal pieces; a third attachment is formed in the same way and the three basalia are called respectively pro-, meso-, and meta-pterygia. By some authors the meta-pterygium and by others the meso-pterygium is regarded as homologous with the axis of the archipterygium.

The pectoral fins of Elasmobranchs vary very much in their mode of attachment. In some of the sharks, including the Notidanidae and _Scyllium_, all three basalia articulate with the pectoral girdle, while in others such as _Cestracion_ the meta-pterygium is excluded. In Rays the propterygium and the meta-pterygium are long and narrow and diverge much from one another; other basalia work their way in between the meso-pterygium and meta-pterygium, and come to articulate with the pectoral girdle. Sometimes they fuse and form a second meso-pterygium. The radiale are greatly elongated and are segmented.

In _Chimaera_ all three basalia are present, but the meso-pterygium is shifted and does not articulate with the pectoral girdle[49].

In _Acipenser_ and _Polyodon_ the pectoral fin is built on the same type as in Elasmobranchs, but becomes modified from the fact that the propterygium is replaced by dermal bone which forms a large =marginal ray=. Extra meso-pterygia are formed in the same way as in Rays.

In _Polypterus_ the pro-and meta-pterygia have ossified while the meso-pterygium remains chiefly cartilaginous; the fin-rays are also chiefly ossified.

In _Amia_, _Lepidosteus_, and certain Teleosteans like _Salmo_, not only the propterygium but the meso-pterygium is almost suppressed by the marginal ray.

In the great majority of Teleosteans a still further stage is reached, the endoskeletal elements, the basalia and radiale are almost entirely suppressed and the fin comes to consist entirely of ossified fin-rays of dermal origin.

In some Teleosteans--_Exocaetus_, a herring, and _Dactylopterus_, a gurnard--the pectoral fins are so enormously developed that by means of them the fish is able to fly through the air for considerable distances. The skeleton of these great fins is almost entirely composed of dermal bone.

PELVIC FIN.

The pelvic fin is almost always further removed from the archipterygial condition, and is in general more modified than is the pectoral. Thus in the Ichthyotomi, while the pectoral fins are archipterygia similar to those of _Ceratodus_, the pelvic fins consist of an axis bearing rays on the postaxial side only, and prolonged distally into a clasper. In Dipnoi however the pelvic fins are very similar to the pectoral. In Elasmobranchs the meso-pterygium is missing, the propterygium is small or absent, and the fin is mainly composed of the meta-pterygium (generally called basi-pterygium) and its radiale. The males in Elasmobranchii and Holocephali have the distal end of the meta-pterygium prolonged into a clasper.

In Ganoids and in Teleosteans the loss of the pelvic girdle causes the pelvic fin to be still further removed from the primitive state. There is always a large basi-pterygium which lies imbedded in the muscles and meets its fellow at its proximal end. In Cartilaginous Ganoids it has a secondary segmentation. Its relation to its fellow is subject to much variation in Teleosteans, sometimes as in the Perch the two are in contact throughout, sometimes as in the Salmon they meet distally as well as proximally, but are elsewhere separated by a space, sometimes as in the Pike and Bony Ganoids they diverge widely. The radiale are articulated to the basi-pterygium. In Cartilaginous Ganoids and _Polypterus_ they are well developed, in other Ganoids and in Teleosteans they are in the main replaced by dermal fin-rays.

In some Teleosteans such as the Cod the pelvic fins have migrated from their usual position and come to be attached to the throat in front of the pectoral fins. Fish with this arrangement are grouped together as =jugulares=.

FOOTNOTES:

[39] The following general works on fishes may be referred to: Bashford Dean, _Fishes, Living and Fossil_, New York, 1895. A. Günther, _An Introduction to the Study of Fishes_, Edinburgh, 1880. A.A.W. Hubrecht and M. Sagemehl, _Fische_ in Bronn's _Classen und Ordnungen des Thierreichs_, Band VI. Leipzig, 1876.

[40] See W.G. Ridewood, _Nat. Sci._ vol. VIII. 1896, p. 380. Full references are there given to the literature of the subject.

[41] See H. Gadow and E.C. Abbott, _Phil. Trans._ vol. 186 (1895) B. pp. 163-221.

[42] C. Hasse, _Zeitschr. wiss. Zool._ LVII. 1893, p. 76.

[43] C. Hasse, _Das natürliche System der Elasmobranchier auf Grundlage des Baues und der Entwickelung ihrer Wirbelsäule_, Jena, 1879 and 1885, and "Die fossilen Wirbel, Morph. Studien I.-IV.," _Morphol. Jahrb. Bd._ II., III. and IV. 1876-78.

[44] See H.B. Pollard, _Anat. Anz._ X. 1894.

[45] T.W. Bridge, "The Cranial Osteology of _Amia calva_," _J. Anat. Physiol. norm. path._ 1876, vol. XI. p. 605. R. Shufeldt, "The Osteology of _Amia calva_," _Ann. Rep. of the Commissioner for Fish and Fisheries_, Washington, 1885.

[46] A. Smith Woodward, _Nat. Sci._ vol. I. 1892, p. 28. Further references are here given on the literature of the subject.

[47] C. Gegenbaur, Ueber das Archipterygium, _Jena Zeitschr. der Wirbelthiere_, 2^e Heft, 1873, vol. 7, and _Morphol. Jahrb._ XXII. 1894, p. 119.

[48] The fins of _Ceratodus_ are very variable, no two being exactly alike. Sometimes even the main axis bifurcates. See W.A. Haswell, _Linn. Soc. N. S. Wales_, vol. VII. 1882.

[49] Some of these views with regard to the homologies of the parts of the fins are not accepted by all anatomists.